Akçay Ö., Altın M., Güler M. A., Acar E.
EUROPEAN JOURNAL OF MECHANICS, A/SOLIDS, cilt.118,106109, sa.106109, ss.1-18, 2026 (SCI-Expanded, Scopus)
-
Yayın Türü:
Makale / Tam Makale
-
Cilt numarası:
118,106109
Sayı:
106109
-
Basım Tarihi:
2026
-
Doi Numarası:
10.1016/j.euromechsol.2026.106109
-
Dergi Adı:
EUROPEAN JOURNAL OF MECHANICS, A/SOLIDS
-
Derginin Tarandığı İndeksler:
Scopus, Science Citation Index Expanded (SCI-EXPANDED), Compendex, INSPEC, MathSciNet, zbMATH
-
Sayfa Sayıları:
ss.1-18
-
Gazi Üniversitesi Adresli:
Evet
Özet
In this paper, the crash performance of a 3D-printed hybrid multi-cell energy absorber (HMCEA) is investigated both experimentally and numerically, and machine learning (ML) based crashworthiness optimization of these tubes is performed. The crash performances of this structure is assessed using two most used metrics: specific energy absorption (SEA) and crush force efficiency (CFE). To predict these crashworthiness metrics, two widely used ML models Gaussian Process Regression (GPR) and Support Vector Regression (SVR) are developed using datasets created from simulation results of a finite element analysis (FEA) model, which is validated through experiments. The ML models are used to predict the effects of various design variables on crash performance and also used in optimization process. When the accuracies of the generated ML models are examined, it is observed that all the constructed models have errors of less than 10%. It is also found that the GPR models with the Matern52 kernel and the squared exponential kernel are the most accurate models for CFE and SEA estimation, respectively. The optimization results show that the optimum HMCEA design provides a 26.1% increase in SEA and a 12.5% increase in CFE compared to the baseline HMCEA design.
In this paper, the crash performance of a 3D-printed hybrid multi-cell energy absorber (HMCEA) is investigated both experimentally and numerically, and machine learning (ML) based crashworthiness optimization of these tubes is performed. The crash performances of this structure is assessed using two most used metrics: specific energy absorption (SEA) and crush force efficiency (CFE). To predict these crashworthiness metrics, two widely used ML models Gaussian Process Regression (GPR) and Support Vector Regression (SVR) are developed using datasets created from simulation results of a finite element analysis (FEA) model, which is validated through experiments. The ML models are used to predict the effects of various design variables on crash performance and also used in optimization process. When the accuracies of the generated ML models are examined, it is observed that all the constructed models have errors of less than 10%. It is also found that the GPR models with the Matern52 kernel and the squared exponential kernel are the most accurate models for CFE and SEA estimation, respectively. The optimization results show that the optimum HMCEA design provides a 26.1% increase in SEA and a 12.5% increase in CFE compared to the baseline HMCEA design.